CONTRIBUTIONS FROM THE MUSEUM OF PALEONTOLOGY
THE UNIVERSITY OF MICHIGAN VOL. 28. N o . 16. PP. 393-416
Sedember 30, 1993
PARTIAL SKELETONS OF ZNDOCETUS RAMANI (MAMMALIA, CETACEA) FROM THE LOWER MIDDLE EOCENE DOMANDA SHALE IN THE SULAIMAN RANGE OF PUNJAB (PAKISTAN)
PHILIP D. GINGERICH, S. MAHMOOD RAZA, MUHAMMAD ARIF, MOHAMMAD ANWAR, AND XIAOYUAN ZHOU
MUSEUM OF PALEONTOLOGY THE UNIVERSITY OF MICHIGAN ANN ARBOR
CONTRIBUTIONS FROM THE MUSEUM OF PALEONTOLOGY Philip D. Gingerich, Director
This series of contributions from the Museum of Paleontology is a medium for publication of papers based chiefly on collections in the Museum. When the number of pages issued is sufficient to make a volume, a title page and a table of contents will be sent to libraries on the mailing list, and to individuals on request. A list of the separate issues may also be obtained by request. Correspondence should be directed to the Museum of Paleontology, The University of Michigan, Ann Arbor, Michigan 48 109-1079. VOLS. 2-28. Parts of volumes may be obtained if available. Price lists are available upon inquiry.
PARTIAL SKELETONS OF INDOCETUS RAMANI (MAMMALIA, CETACEA) FROM THE LOWER MIDDLE EOCENE DOMANDA SHALE IN THE SULAIMAN RANGE OF PUNJAB (PAKISTAN)
PHILIP D. GINGERICH, S. MAHMOOD RAZA~,MUHAMMAD ARIFI, MOHAMMAD ANWAR1, AND XIAOYUAN ZHOU
Abstract-Indocetus ramani is an early middle Eocene archaeocete described by Sahni and Mishra from Kutch in western India. The type and only specimen known previously is a partial skull. New specimens from the marine lower middle Eocene Domanda Shale of the Sulaiman Range in Pakistan show that Indocetus retained a long neck; a robust, solidly-fused sacrum composed of a minimum of four vertebrae; a pelvis with a large, deep acetabulum; a robust femur (incompletely preserved); and a tibia of normal mammalian proportions, including distal articular facets for a broad astragalus. The pelvis from the Indocetus ramani type locality that Sahni and Mishra described as Protosiren fraasi and the solidly fused sacrum from the type locality that they described as cf. Moeritheriid almost certainly belong to Indocetus ramani as well. Indocetus, placed in the new subfamily Indocetinae, is an important intermediate stage in the early evolution of Cetacea. It combines an advanced Protocetus-like archaeocete skull with a more primitive postcranial skeleton retaining the long neck and well developed hind limbs of a land mammal.
INTRODUCTION Protocetus atavus from the middle Eocene of Egypt was, for a long time, the oldest and most primitive cetacean known in the fossil record. Protocetus is represented by a skull with teeth, and by several isolated vertebrae. Primitive features include retention of a full dental formula of 3.1.4.3, retention of protocones on upper molars, and retention of a distinct sacrum with iliosacral articulations at the ends of the transverse processes (Fraas, 1904). Advanced features include broad flat frontals and an elongated, anteriorly-protruded palate and dentition, giving the skull the characteristic shape of an archaeocete. Protocetus is found in the Lower Building Stone Member of the marine Mokattam Formation, which is middle Lutetian in age
' ~ e o l o ~ i cSurvey al of Pakistan, 84 H-811, Islamabad
P. D. GINGERTCH ET AL. 70" OO'E 31" N
70" 30'E
..................
30" N
Whale-bearing Eocene Domanda Shale Habib Rahi Limestone Fossil Localities
Dry Tributaries of River lndus
70" OO'E
FIG. I-Map
70" 30'E
showing outcrop pattern of Domanda Shale and early middle Eocene fossil localities discussed in the text. Prospecting in 1977 and 1981 was concentrated on exposures on the eastern flank of the Sulairnan Range in the vicinity of Rakhi Nala. Our 1992 expedition investigated exposures in Bozmar Nadi near the northern end of the Zinda Pir anticlinorium. Both areas are richly fossiliferous, and much intervening outcrop remains to be explored.
IhDOCETUS RAMANI FROM PAKISTAN
395
(probably planktonic foraminifera1 zone P11, ca. 46 million years before present [Ma]), deposited in a marine shallow shelf environment of the ancient Tethys Sea (Gingerich, 1992). In recent years, new protocetid and other primitive cetaceans have been described from eastern Tethys in India (Sahni and Mishra, 1972, 1975; Kumar and Sahni, 1986) and in Pakistan (Dehm and Oettingen-Spielberg, 1958; West, 1980; Gingerich and Russell, 1981, 1990; Gingerich et al., 1983; Thewissen and Hussain, 1993), but none of these specimens includes significant postcranial remains. Experience in Egypt shows that many of the most interesting evolutionary changes in early whale evolution involved the postcranial skeleton (Gingerich et al., 1990). Indocetus ramani is the most Protocetus-like of all of the eastern Tethyan cetaceans. This was described by Sahni and Mishra (1975) from the marine lower middle Eocene Chocolate or Shell Limestone near Harudi in Kutch in western India. A correlative Pakistan formation is the marine lower middle Eocene "lower chocolate clays" of Eames (1952b,c), Domanda Shale Member of the Khirthar Formation (Hemphill and Kidwai, 1973; Wells, 1984; Geological Survey of Pakistan, 1986; Qureshi et al., 1987), or Sirki Member of the Khirthar Formation (Cheema, Raza, and Ahmad, in Shah, 1977). The "lower chocolate clays" are a unit on the order of 300 m thick, mappable over a large area, that Gingerich et al. (1979) treated as a full formation: the Domanda Formation (erroneously attributing the rise to formational status to Cheema et al.). Here we will refer to the "lower chocolate clays" simply as Domanda Shale, leaving aside the question of formational status. The Domanda Shale is extensively and well exposed on the eastern flank of the Sulaiman Range and in the parallel Zinda Pir anticlinorium farther to the east (Fig. 1). The Domanda Shale in Pakistan has now yielded isolated cranial and postcranial elements of mammals from five localities of interest here (Table 1). Most of the isolated elements found in 1977 and 1981 could not be identified when they were found, but the cranium from Debrha, GSP-UM 1853, was referred to Indocetus on the basis of size when it was first studied. More careful comparison with a cast of the holotype and comparison of associated postcranial elements confirm this identification (see below). During our 1992 expedition, one of us (M. Arif) found two fragmentary partial skeletons, GSP-UM 3009 and 3015, each including associated cranial and postcranial remains of an archaeocete cetacean. These were found 20 m apart at the same stratigraphic level in brown clays of the upper Domanda Shale near the mouth of Bozmar Nadi (valley) just south of Bor or Sohr Nala (Dhodak quadrangle, 39 J/5; Geological Survey of Pakistan, 1986) in the foothills of the Sulaiman Range northwest of Taunsa in Punjab Province. Both have Indocetus-like vomers and both preserve identical axis vertebrae, showing that two conspecific skeletons are represented (a third vomer-like bone may indicate that fragmentary parts of two skulls are represented in GSP-UM 3015). The new specimens are sufficient in themselves to document several distinctive skeletal characteristics of the archaeocete represented, and they permit one skull and several vertebrae and other postcranial elements collected in previous years to be referred to Indocetus ramani, revealing much more about this important genus and species than was known before. ABBREVIATIONS Institutional abbreviations used in this paper are as follows: GSP-UM
LUVP SMN YPM-PU
Geological Survey of Pakistan-University of Michigan collection, Islamabad - Lucknow University vertebrate paleontology collection, Lucknow - Staatliches Museum fiir Naturkunde, Stuttgart - Princeton University collection at Yale Peabody Museum, New Haven -
P. D. GINGERICH ET AL. TABLE 1-Localities in Pakistan yielding cranial and postcranial remains of Indocetus ramani. All localities are in the eastern foothills of the Sulaiman Range, southwestern Punjab Province (Fig. 1). Field no.
Specimen
Description
1. RN-2, south side of Rakhi Nala (29'56'N latitude, 70°7'E longitude) on D. G. Khan series map 39W1 77-0 15 GSP-UM 9 Proximal caudal vertebrae in articulation (2) 77-020 GSP-UM 15 Maxilla fragment Larger specimens questionably referred to I. ramani: 77-016 GSP-UM 11 Sacrum 77-0 17 GSP-UM 12 Vertebral fragments (possibly part of GSP-UM 11) 77-01 9 GSP-UM 14 Atlas (2 pieces) 2. RN-3, north and south sides of Mihal Nala (30°3'N latitude, 70°9'E longitude) on map 39514 (Gulu Sheikh) 77-030 GSP-UM 19 Cervical vertebra (C5 or C6, centrum w. transverseprocesses) 77-03 1 GSP-UM 77 Sacrum 77-032 GSP-UM 20 Lumbar vertebra (ant. part of centrum) Larger specimen questionably referred to I. rainani: 77-029 GSP-UM 18 Acetabulum of pelvis 3. RN-4, 1.6 km north of Rakhi Nala (29'58'N latitude, 70°7'E longitude) on map 39Kl1 Larger specimen questionably referred to I. ramani: 77-050 GSP-UM 24 Thoracic vertebra (centrum) 4. RN-5, Debrha (30°1'N latitude, 70°8'E longitude) on map 39514 (Gulu Sheikh) 8 1-262 GSP-UM 1853 Cranium 8 1-276 GSP-UM 1854 Lumbar vertebra (centrum)
5. BN-1, near mouth of Bozmar Nadi (30°47'N latitude, 70°27'E longitude) on map 39515 (Dhodak) 92-013 92-015
GSP-UM 3009 GSP-UM 3015
Cranial fragments, vertebrae, sacrum, acetabulum of pelvis Cranial fragments, vertebrae, proximal femur, tibia
CRANIUM The type specimen of Indocetus ramani, LUVP 11034, is a partial cranium from the Chocolate o r Shell Limestone near Harudi in Kutch, western India. This was described by Sahni and Mishra (1975) and illustrated in their plate IV, figures 1-3. The type is a skull in two pieces, one including the frontal shield and the other including the right occiput and right auditory bulla. Sahni and Mishra mentioned cranial elongation, a lateral profile with the frontal meeting the sagittal crest at an obtuse angle, a postorbital process of moderate size, a massive typically whale-like bulla, and retention of a well developed protocone on upper molars as characteristics of the genus. The new skull found in 1981, GSP-UM 1853, is in several pieces, of which the most important is the neurocranium with the occiput and right auditory bulla (Figs. 2 and 3). The right frontal is attached, but large parts of the sagittal and nuchal crests are missing. A second piece includes the left frontal and much of the rostrum and palate, extending as far forward as the alveoli for left and right PI. A third piece includes the anterior parts of both premaxillae, with alveoli for all three incisors. Measurements of the new skull are compared with those of the holotype in Table 2.
INDOCE77J.9 RAMAM FROM PAKISTAN
397
FIG. 2-Right exoccipital and squamosal of skull of Indocetus ramani, GSP-UM 1853, from Debrha. Note flat posterior surface of exoccipital (arrow) sirnilar to that in the type of Indocetus ramani. Specimen is shown in right lateral view. See Figure 3 for orientation and scale. Abbreviations: b, auditory bulla; ex, exoccipital; occ, occipital condyle; par, parietal; pet, petrosal; pgp, postglenoid process.
The occiput and auditory bulla of GSP-UM 1853 resembles that of LUVP 11034 very closely. These are distinctive among known protocetids in having a relatively flat posterior surface of the exoccipital (Fig. 2). The foramen magnum opens more posteriorly than ventrally. Another important feature of the new skull is retention of stone endocasts of the left and right narial passages, showing that the external nares opened above the canine alveoli, anterior to the front of preserved P' alveoli and posterior to preserved alveoli for 13 in the premaxillae. These stone endocasts envelope the vomer, which is a large, long, sharply keeled bone that is now exposed ventrally in the midline of the skull. The narial passages and vomer are normally covered ventrally by palatines, but these are broken away in GSP-UM 1853. Isolated vomers of GSP-UM 3009 and 3015 are illustrated in Figure 4. The right auditory bulla in GSP-UM 1853 measures 62 mm in length anteroposteriorly and 45 mm in breadth transversely. Sahni and Mishra (1975, p. 19) gave these measurements as 70 mm and 45 mm, respectively, in the holotype. The new bulla is not well preserved, but it does confirm that Indocetxv had large, dense, typically whale-like tympanic bullae. Alveoli show the upper incisors and P' to have been single-rooted. was double-rooted. P was double-r~oted,with a broad posterior root. appears to have been three-rooted, and M' and M2 were clearly three-rooted. M3 was double-rooted, or possibly single-rooted (it is not possible to determine whether the roots for M3 diverged). The presence of an alveolus for M~ is important because it is not possible to determine from the type skull whether this tooth
TABLE 2-Measurements of the new cranium of Indocetus ramani, GSP-UM 1853, from Debrha in Pakistan, compared to the holotype of Indocetus ramani, LUVP 11034, from Harudi in Kutch, western India (measurements of holotype from Sahni and Mishra, 1975, p. 20), and to the holotype of Protocetus atavus, SMN 11084, from Gebel Mokattam in Egypt (measurements from Fraas, 1904, p. 203).
Indocetlts ramani
Protocetus atavus
LUVP 11034 GSP-UM 1853 (holotype)
SMN 11084 (holotype)
Condylobasal length:
62 cm*
---
58 cm
Length of the P3-M2 series
11.2 cm
10 cm
---
Width of skull at level of anterior root of
8.6 cm
8 cm
---
Width across supraorbital processes of frontal shield:
21 cm
20 cm
16.7 cm
Width across exoccipitals:
22 cm
22 cm
---
Width across zygomatic arches:
26 cm'
---
24 cm
'Estimate
was retained or not, and the new skull shows clearly that M3 was still present in Indocetus (M3 is lost in some later archaeocetes). Measurements of roots and alveoli of upper teeth in the new skull are compared to measurements of alveoli of upper teeth in the holotype in Table 3. These are generally similar, and differences may be attributable to the difficulty of measuring alveoli consistently. Judging from alveoli, the dentition of Indocetus was much like that of Protocetus. The only tooth of Indocetus known to date is the partial crown of M' illustrated by Sahni and Mishra (1975) in their plate IV, fig. 3. This is distinctive in having a lingual cingulum surrounding a broad and flatly worn protocone. DENTARY Dentaries were not found with the type specimen of Indocetus ramani, and no recognizable dentaries have been found in subsequent collections either. However, there is one Indian specimen, LUVP 11003, that is too large to belong to any of the taxa to which it has been referred. This dentary measures 60 to 63 mm deep beneath alveoli for P, and M,, respectively (Sahni and Mishra, 1972, p. 493). LUVP 11003 was first described and illustrated by Sahni and Mishra (1972, p. 493, plate 97, figs. 6-7), who identified it as Protocetus sloani (type specimen LUVP 11002). Sahni and Mishra (1972, p. 492) implied that LUVP 11002 and 11003 were part of a collection of cetacean remains from the gypsiferous clays 3 km southeast of Baranda in Kutch. Later, Sahni and Mishra (1975, p. 21) listed LUVP 11002 and 11003 as part of a collection from the Shell Limestone and from ossiferous gypseous shales at Harudi (which is 3 km southeast of Baranda). Finally, Kumar and Sahni (1986, p. 341) indicated that LUVP 11002 came from the Chocolate Limestone at the Harudi fossil locality. Kumar and Sahni did not mention LUVP 11003, but this shares the same good preservation with bones from the Chocolate Limestone, and it is reasonable to suspect that LUVP 11003, like LUVP 11002, came from the Chocolate Limestone at Harudi. If so, then LUVP 11003 is from the same locality and horizon as the
IADOCETUS RAMAM FROM PAKISTAN
399
FIG. 3-Craniurn of Indocerus ranuzni, GSP-UM 1853, from Debrha. A and B, dorsal and right lateral views. Note elongated rostrum and braincase, broad frontal shield V) above orbits, and cast of nares (n) extending forward to open above canine (missing region of rostrum). Supraoccipital nuchal crest, zygomatic arches, and all teeth are also missing. Scales are in cm. Measurements are given in Tables 2 and 3.
type specimen of Indocetus ramani, with which it agrees much more closely in size and form than it does with other dentaries of Protocetus (now Remingtonocetus) sloani (see Kumar and Sahni, 1986, p. 341). The importance of LUVP 11003 is in showing that the mandibular symphysis in Indocetus ramani was not fused back as far as P,. This specimen has the potential, with further cleaning, to show the size of the mandibular foramen or canal in Indocetus, which might indicate the degree of development of underwater hearing in the genus. Thewissen and Hussain (1993) stated that Pakicetus is the only cetacean with a small mandibular foramen. A good cast of LUVP 11003 provided by Kumar and Sahni gives no indication of enlargement of the mandibular canal, but this needs to be checked on the original specimen. VERTEBRAL COLUMN GSP-UM 3009 and 3015 are important because they preserve, respectively, nine and fifteen vertebral centra from all parts of the vertebral column. These are not as informative as more or better preserved vertebrae would be, but they do reveal a number of distinctive features of Indocetus rarnani that were not known previously.
P. D. GINGERICH ET AL.
400
FIG. 4--Elongated vomers of Indocetus ramani from Bozmar Nadi. A, central portion of GSP-UM 3009. B, posterior tip of GSP-UM 3009. C, posterior portion of GSP-UM 3015. All are shown in ventral view. Overlap in preserved parts shows that two specimens are represented. Scale is in cm. TABLE 3-Comparison of measurements (mm) of the roots or alveoli (in parentheses) of cranial teeth in Indocetus ramani, GSP-UM 1853, from Debrha in Pakistan, with alveoli (in parentheses) in the holotype LUVP 11034 from Harudi in India. GSP-UM 1853 Tooth position
Length
Width
LUVP 11034 (holotype) Length
Width
Cervical vertebrae-GSP-UM 14 is the only atlas vertebra known from the upper Domanda Shale (Fig. 5A). It articulates reasonably well with GSP-UM 1853, but seems large and surprisingly advanced in morphology (with flat axis facets like those of Egyptian Prozeuglodon), and it is here questionably referred to I. ramani. GSP-UM 14 is broken, with only the left side preserved. The anterior articulation for the left occipital condyle spirals obliquely forward and upward. The posterior articulation is kidney-shaped, measuring 45 mm in maximum length and about 31 mm in maximum breadth; it is distinctive in being almost perfectly flat, and it is not clear how this could have articulated with an axis vertebra like that
INDOCETUS RAMAM FROM PAKISTAN
401
FIG. 5-Cervical vertebrae from Rakhi Nala and Mihal Nala referred to Indocetw ramani. A, portion of left half of atlas vertebra, GSP-UM 14 from RN-2, in cranial view. Note curved anterior articular surface for left occipital condyle of skull. Posterior articular surface is kidney-shaped and almost flat. B, centrum of C5 or C6, GSP-UM 19 from RN-3, in cranial view. Note weak hypapophysis (hyp) and downwardly-angled transverse processes. Scale is in cm.
FIG. 6-Centra of cervical vertebrae of Indocetus ramani from Bozmar Nadi. All are missing neural arches. A, centrum of axis vertebra, GSP-UM3015, with dens and very large hypapophysis (Iryp). B, centrum of middle cervical vertebra, GSP-UM 3015, with large hypapophysis. C, centrum of middle or posterior cervical vertebra, GSP-UM 3015, with reduced hypapophysis. D, centrum of posterior cervical vertebra, GSP-UM 3015, lacking hypapophysis. E, poorly preserved centrum of axis vertebra, GSP-UM 3009, with broken dens and eroded hypapophysis. Recovery of two axis vertebrae shows that two specimens are represented. All are shown in right lateral view. Scale in cm. associated with GSP-UM 3015, which appears to have had a more oblique articulation. This specimen indicates either that the axis of I. ramni had broad flat anterior articular surfaces or, alternatively, the presence of a larger, more advanced cetacean in the upper Domanda Shale. GSP-UM 14 has a large arterial foramen 7.5 mm in diameter perforating the transverse process. The functional length of the atlas is 15.5 mm, which is the closest possibly approximation of the occipital condyles of the skull to the anterior articular surface of the axis vertebra (other vertebral measurements are given in Table 4). There are two axis vertebrae, one preserved with GSP-UM 3009 (Fig. 6E) and the other with GSP-UM 3015 (Fig. 6A). The former is badly eroded, and adds nothing beyond what can be seen in GSP-3015. The latter is distinctive in having a relatively large projecting odontoid process, a long centrum, and an extraordinarily large hypapophysis.
402
P. D. GINGERICH ET AL.
FIG. 7-Thoracic, lumbar, and caudal vertebrae of Indocetus ramani, GSP-UM 3015, from Bozmar
Nadi. A, centrum of anterior thoracic vertebra. B, centrum of posterior thoracic vertebra. C, centrum of lumbar vertebra. D, anterior caudal vertebra with neural arch, pre- and postzygapophyses,. and short transverse processes. Specimens are shown here in right lateral view (compare Fig. 8). Scale is in cm. Other cervical vertebrae found with GSP-UM 3015 (Fig. 6B-D) have centra ranging in length from 36.0 to 42.6 mm, averaging about 40 mm, which is long for an archaeocete of this size. The centra have obliquely-set articular surfaces (relative to the cranial-caudal axis of the body), like those of Pachyaena ossfiaga (Zhou et a]., 1992), which indicates that the head was habitually carried farther forward and higher relative to the rest of the body than was true in more advanced archaeocetes. As in Pachyaena, the cranial surface of the centrum is more flattened in anterior view, with a distinct groove crossing it transversely, and the caudal surface of the centrum is more rounded and slightly concave. Finally, these show progressive reduction in the size of the hypapophysis in more posterior vertebrae. One specimen, GSP-UM 19, is a cervical with large, downward-turned transverse processes indicating that it is probably C5 or C6 (Fig. 5B). This has large pedicles for a strong neural arch dorsally, and a broad hypapophysis ventrally. The arterial canals perforating the base of each transverse process measure about 6 mm in diameter. 7horacic vertebrae-Thoracic vertebrae are difficult to position in the vertebral column unless pedicles, neural spines, and rib articulations are well preserved. None of the vertebrae available here preserve neural arches or rib articulations. Anterior thoracics are identified on the basis of their short and deep centrum (Fig. 7A), which tapers laterally toward the rib articulations and neural arches (Fig. 8A). Posterior thoracics have longer centra, and retain some evidence of articulating ribs. Measurements of thoracic centra are listed in Table 4, which shows that anterior thoracics of I. ramani averaged about 40 mm in length, while posterior thoracics averaged about 47 mm in length. Lumbar vertebrae-Lumbar vertebrae of Indocetus have relatively long centra that are oval to symmetrically kidney-shaped in cross-section (Figs. 7B,C and 8B,C). These average about 54 mm in centrum length (Table 4). Sacrum-One of the most interesting bones found with GSP-UM 3009 is a partial sacrum in two pieces. The larger piece has the centra of the second and third sacral vertebrae (S2 and S3) solidly fused, with co-ossified transverse processes (pleurapophyses) forming a broad marginal shelf for muscle attachment (Figs. 9B, IOA). This shelf encloses left and right dorsal and ventral sacral foramina that carried spinal nerves. The sacrum retains a stone endocast of the neural canal, which is roughly triangular in cross-section, measuring about 17.5 mm high
IDOCETUS RAMANI FROM PAKISTAN
403
FIG. 8-Thoracic, lumbar, and caudal vertebrae o f Indocetus ramani, GSP-UM 3015, from Bozmar Nadi. A, centrum o f anterior thoracic vertebra. B, centrum o f posterior thoracic vertebra. C, centrum o f lumbar vertebra. D, anterior caudal vertebra with neural arch, pre- and postzygapophyses, and short transverse processes. Specimens are shown here in anterior view (compare Fig. 7). Scale is in cm.
and 22.5 mm wide at the base. This stone endocast shows clearly how left and right spinal nerves angled backward from the spinal cord and then branched to send off dorsal and ventral rami through the sacral foramina. The second sacral vertebra preserves the posterior part of the right auricular surface for articulation with the pelvis, which is how it can be identified with certainty as S2. The smaller piece of sacrum in GSP-UM 3009 includes a large part of the auricular surface of the left side (Fig. 9A), which was originally part of both S1 and S2. The auricular surface of the sacrum is large relative to the rest of the sacrum, indicating the substantial articulation with the pelvis characteristic of land mammals. There is a sharp break at the posterior end of S3 (Fig. 9B), which appears not to have been firmly co-ossified with S 4 , but the broad marginal shelf formed by sacral pleurapophyses extends to the back of S3, with a roughened surface showing where the pleurapophysis of S3 articulated with the pleurapophysis of S 4 , meaning that the full sacrum included a minimum of four sacral vertebrae. Finally, the neural arch in GSP-UM 3009 is partially preserved on S3 but not on S2. Recovery of a fused sacrum with GSP-UM 3009 indicates that sacra found in the Domanda Shale in 1977 are probably cetacean as well. The best preserved of these is GSP-UM 77 from Mihal Nala (RN-3), mentioned by Gingerich et al. (1979, p. 109) and illustrated here in Figures 9C and 10B. This is virtually identical to GSP-UM 3009 in size (Table 4) and in the form of comparable parts, which include parts of the left auricular surface and the body of S2. GSP-UM 77 preserves the first and second sacral centra (S1 and S2) solidly co-ossified. Parts of both left and right auricular surfaces are preserved (and the sacrum is about 95 mm in width across what remains of these), but neither is as well preserved as the left auricular surface of
FIG. 9-Sacra of Indocetus ranlani from Bozmar Nadi and Mihal Nala. A, left auricular surface of the transverse process of the sacrum in GSP-UM 3009. This auricular surface indicates a large area of articulation with the pelvis. B, fused sacral vertebrae S2 and S3 in GSP-UM 3009. Auricular surface of transverse process is partially preserved, and the broad marginal shelf formed by fusion of pleurapophyses of both vertebrae is well preserved on the right side of this sacrum. S3 preserves part of neural arch, and an endocast of the neural canal. The marginal shelf ends abruptly at the posterior end of S3 (arrow), showing that this sacrum had a fourth vertebra with well-formed pleurapophyses that was less solidly fused to the others. C, fused sacral vertebrae S 1, S2, and part of S3 in GSP-UM 77. Left and right auricular surfaces of the transverse processes of S1 and S2 are partially preserved. Pre- and postzygapophyses, pleurapophyses, and vertebral centra are all solidly fused. Specimensare shown in dorsal view; anterior is to the right (compare Fig. 10). Scale is in cm.
GSP-UM 3009. Each auricular articulation runs the length of S1 and most of S2, where these merge smoothly into the broad marginal shelf formed by co-ossified pleurapophyses. Dorsal and ventral sacral foramina are well preserved perforating this shelf where S1 joins S2 and where S2 joins S3. Prezygapophyses and postzygapophyses joining S1 to S2 and joining S2 to S3 are solidly co-ossified, and each set of zygapophyses surrounds deep paired plts for tendinous muscle origination. Neural spines are present, but these are not well preserved. In cranial view, the anterior articular surface of the centrum of S1 is oval in outline and compressed dorsoventrally, with a height-to-width ratio of 0.61 or 1.65, depending on how this is expressed (based on measurements in Table 4). The lateral margins of the anterior articular surface of S1 are defined by grooves for spinal nerves, which separate this articular surface from the impinging transverse processes. The neural canal is well preserved. As in GSP-UM 3009, this is triangular in cross-section. The canal measures 27 mm wide at the base, and it is 15.5 mm high in the midline.
IM)OCETC/S RAMAM FROM PAKISTAN
405
FIG. 10-Sacra of Indocetus ramani from Bozmar Nadi and Mihal Nala. A, fused sacral vertebrae S2 and S3 in GSP-UM 3009, with part of auricular surface (arrow) and the broad marginal shelf formed by fusion of pleurapophyses showing on the right side. B, fused sacral vertebrae S 1, S2, and part of S3 in GSP-UM 77. Left and right auricular surfaces of the transverse processes are partially preserved. Specimens are shown in anterior view (compare Fig. 9). Scale is in cm. Taken together, GSP-UM 77 and GSP-UM 3009 indicate that the sacrum of Indocetus ramani was long (ca. 210 to 220 mm long), relatively narrow (estimated at ca. 115 mm wide), and massive, with large auricular processes for firm articulation with the pelvis. A third partial sacrum, GSP-UM 11 from Rakhi Nala locality RN-2, preserves S1 and part of 52. This may represent Indocetus ramani, but it is even larger than GSP-UM 77 and GSP-UM 3009. It differs somewhat in shape (a difference possibly exaggerated by a different pattern of breakage), and GSP-UM 11 may represent a different and slightly larger archaeocete. Caudal vertebrae-The best preserved caudal vertebrae are GSP-UM 3015 (Figs. 7D and 8D) from Bozmar Nadi and GSP-UM 9 (two caudals preserved in articulation) from Rakhi Nala locality RN-2. GSP-UM 3015 has a well preserved centrum and the neural arch intact. The centrum is distinctive in having very short, robust, transverse processes, indicating that this is one of the anteriormost caudals, and in having a pair of well developed chevron (hemal arch) articulations on the posteroventral margin of the centrum. The dorsal part of the neural arch arises from pedicles extending along the anterior two-thirds of the dorsal surface of the centrum, and this bears large tlat prezygapophyses facing inward and oriented at an angle of about 45" to the sagittal plane. These extend well forward of the anterior surface of the centrum. Large flat postzygapophyses face outward and these too are oriented at an angle of about 45". These extend just past the posterior surface of the centrum. By comparison, the neural spine appears small and weak (some of the dorsal part of this is missing). The neural canal is triangular in cross-section like that of the sacrum; this measures about 18 mm wide at the base and 13 mm high in the midline. The more anterior of the two anterior caudal vertebrae in articulation in GSP-UM 9 is very similar to the anterior caudal of GSP-UM 3015 described above, differing in having a smaller neural canal and thinner transverse processes. The more posterior caudal vertebra of GSP-
P. D. GINGEIUCH ET AL.
406
TABLE 4-Measurements (mm) of vertebral centra of Indocetus ramani based on specimens from the upper part of the Domanda Shale. Centrum lengths of sacral vertebrae are measured to and/or from centers of sacral foramina; width and height are given for anterior face of S1 only. Dittos mark separate centra preserved in articulation. Centrum GSPUM
Vertebra
14 3009 3015 3015 3015 19 3009 3015
Atlas (Cl; Fig. 5A, quest. ref.) Axis (C2, eroded; Fig. 6E) Axis (C2; Fig. 6A) Cervical (C3?; Fig. 6B) Cervical (C4 or C5; Fig. 6C) Cervical (C5 or C6) Cervical (C7?) Cervical (C7; Fig. 6D)
Locality
Length
Width
Height
BN-1 BN-1 BN-1 RN-5
48.7 53.3 55.1 60.2
53.6 60.0 57.3 57.5
40.4 36.8 36.7 39.7
BN-1
51.7 55.7 56.5 51.5 65.8
-----
-----
65.1
39.5
-----
45.3
Anterior thoracic Anterior thoracic Ant. thoracic (quest. ref.) Anterior thoracic Ant. thoracic (Fig. 7A, 8A) Anterior thoracic Anterior thoracic Posterior thoracic Posterior thoracic Posterior thoracic(?) 3015 3015 3015 1854
Lumbar (Figs. 7B, 8B) Lumbar (Figs. 7C, 8C) Lumbar Lumbar
3009
Sacral Sacral Sacral Sacral Sacral
"
77 " 11 3009 3009 3015 3015 9 "
(S2; (S3; (Sl; (S2; (Sl,
Fig. 9B, 10A) Fig. 9B, 10A) Fig. 9C, 10B) Fig. 9C, 10B) questionably ref.)
Proximal caudal Proximal caudal Prox. caudal (Figs 7D, 8D) Prox. caudal (dorsovent. flat.) Proximal caudal Proximal caudal
RN-3 RN-2 BN-1 BN-1 BN-1 BN-1 RN-2
55.6 48.3 48.2 49.2 48.6 44.5
---
58.9 49.0
---
--38.8 40.3
---
---
-----
-----
'Estimate
UM 9 is similar to the vertebra in front of it, but the neural spine slopes forward as if to articulate with the preceding postzygapophyses, and the postzygapophyses on this vertebra are much reduced in size. The remaining caudal vertebrae are centra that add little to our understanding of tail structure in Indocetus.
IhDOCETUS RAMANI FROM PAKISTAN
I
I
0
I
I
10
I
I
20
I
I
I
30
I
40
I
I
50
407
I
I
60
I
I
70
I
Vertebral number (position in column)
FIG. 11-Averaged length-of-vertebrae profile of partially marine middle Eocene lndocetus ramani (open circles) compared to length-of-vertebrae profile of the terrestrial North American early Eocene mesonychid Pachyaena ossifraga (closed-and-open circles; from Zhou et al., 1992). This shows I. ranuni to have been a little larger than P. ossifraga. Cervical vertebrae of, Indocetus are proportionally as long as those of terrestrial Pachyaena, not shorter like those of later archaeocetes and other cetaceans. Indocetus had a sacrum with at least four sacral centra, while Pachyaena had a sacrum with three sacral centra. The greatest difference in vertebral columns of the two genera appears to be the relatively longer (and heavier) sacrum and proximal caudal vertebrae of Indocetus, suggesting that aquatic locomotion in Indocetus may have been driven by a tail powered by strong dorsal and ventral sacrococcygeal muscles originating from the lumbars, sacrum, and anterior caudal vertebrae. The number of cervical, thoracic, lumbar, and caudal vertebrae in Indocetus is not yet known; values plotted here are averaged from Table 4.
Vertebral size and proportions-It is not possible to say very much about the length or structure of the vertebral column in Indocetus ramani without more complete specimens. However, assuming that this taxon had 7 cervical vertebrae, 14 thoracics, 5 lumbars, 4 sacrals, and a tail of moderate length, we can construct a length-of-vertebrae profile based on the average length of vertebrae at each position documented in Table 4. This length-of-vertebrae profile is compared with that of the mesonychid land mammal Pachyaena ossifaga in Figure 11. The plot shows that I. ramani was uniformly a little larger than P. ossifraga, and it shows that the neck of Indocetus had the length typical of a land mammal of its size. Later and more advanced aquatic archaeocetes and other cetaceans have much shorter cervical vertebrae for their size than those described here for Indocetus. The largest consistent discrepancy between Indocetus and Pachyaena in this vertebral length profile is in the sacral region, where Indocetus has a much longer sacrum than Pachyaena, with longer sacral centra and a minimum of four sacral vertebrae instead of three, and in the proximal caudal region, where Indocetus has longer vertebrae (both absolutely and relatively) than Pachyaena. HIND LIMB Pelvis-GSP-UM 3015 includes much of a right pelvis (Fig. 12B). The acetabulum is large and deep, with a diameter of 40 mm around the margin of the lunate surface and a depth of
408
P. D. GINGERICH ET AL.
FIG. 12-Partial right pelves of archaeocete cetaceans from the Domanda Shale of Pakistan. A, Acetabulum questionably referred to Indocetus, GSP-UM 18from Mihal Nala, in right lateral view. B, right pelvis of Indocetus ramani, GSP-UM 3015 from Bozmar Nadi, in right lateral view. Scale is in cm. about 17 mm (measured to the deepest part of the lunate surface, not into the acetabular fossa). The acetabular fossa opens posteroventrally, directly opposite the ilium. The dorsal surface of the pelvis above the acetabulum is widest transversely above the anterior edge of the acetabulum, where it measures 35 mm. A short section of the body of the ilium is preserved; this is oval in cross-section, with the longest diameter of the oval being 36.5 mm and the shortest diameter of the oval being 20.5. The rest of the body of the ilium is missing, including all of the sacral articular surface. A short section of the body of the ischium is also preserved, which is D-shaped in cross-section, with the longest axis of the D being the medial surface running from the dorsal surface to the obturator foramen. This measures 36.5 mm. A shorter axis of the D, perpendicular to the longest axis, measures 16.2 mm. The broken base of the pubic ramus is present, but nothing is preserved of the ramus itself. The obturator foramen was clearly large, but this is not sufficiently preserved to permit measurement. There is a raised ridge of bone running anteriorly from the dorsal margin of the acetabulum on the lateral side of the ilium (corresponding to an iliopectineal tubercle?), and this is separated from the anterior margin of the acetabulum by a distinct shallow fossa measuring about 2.4 mm in length and 1.2 mm in width. The acetabular region of another right pelvis, GSP-UM 18 (Fig. 12A), was found at Mihal Nala (RN-3) in 1977. This is less well preserved than GSP-UM 3015, and its shallow acetabulum makes reference to Indocetus questionable. The acetabulum is about 40 mm in diameter measured around the margin of the lunate surface, but this is only about 10 mm deep (measured as before). Femur-The proximal end of the right femur is preserved with GSP-UM 3015 (Fig. 13A). This is large and heavily built, with a massive greater trochanter, a deep trochanteric fossa, and a strongly developed intertrochanteric crest. The femoral head and neck are broken away and missing, as is the lesser trochanter. To judge from the size of the acetabulum of GSP-UM 3015 described above, the femoral head must have been about 40 mm in diameter. The femoral shaft where it is broken 10 cm below the top of the greater trochanter, probably somewhat above midshaft, measures 34.7 x 27.7 mm in diameter. The full length of the femur cannot be measured, of course, but if it was as long as the tibia then it was probably something on the order of 25 cm in length. Tibia-Both proximal and distal ends of the right tibia are preserved with GSP-UM 3015 (Fig. 13B). These were originally found in separate concentrations of bone some 10 m apart in the field and regarded as parts of two different individuals, but later a piece of midshaft
INDOCETUS RAMAM FROM PAKISTAN
409
FIG. 13-Hind limb elements of Indocetus ramani, GSP-UM 3015, from Bozmar Nadi. A, proximal end of right femur with large greater trochanter kt), in anterior view (head is missing). B and C, proximal and distal ends of right tibia, in anterior view. D, right tibia of Pacliyaena os,nJ;aga, YPMPU 14708, in anterior view for comparison. Scale is in cm.
discovered with the distal end was found to fit onto the proximal end, demonstrating that both concentrations of bone belong to a single individual animal. The tibia of Indocetus ramni compares closely in size and form to the tibia of Pachyaena ossiJiaga (Fig. 13C). The proximal end is slightly eroded, but medial and lateral condylar surfaces (plateaus) are well defined and these appear to have been about equal in height, like those of Pachyaena. The tibial tuberosity for insertion of the patellar ligament appears to have been less strongly developed than in Pachyaena, but the tibial or cnemial crest curves medially and downward as in Pachyaena making the upper shaft triangular in cross-section. The proximal end measures 57 mm in breadth across the condylar surfaces and approximately 44 mm anteropsteriorly across the tibial tuberosity. The distal end of the tibia is missing much of its surface bone, but it flares to a width of approximately 43 mm distally, and the
410
P. D. GINGERICH ET AL.
distal articular surface is well preserved. This consists of two shallow concavities positioned side-by-side, separated by a low rise of bone, indicating that the proximal articular surface of the body of the astragalus (not found) was shallowly trochleated like that of Pachyaena. This articulation is very different than the tibio-astragalar articulation in later Basilosaurus (Gingerich et a]., 1990) where these two bones join along a single simple surface. The tibia of Indocetus is very different in being larger and longer, both relatively and absolutely, and it is different in lacking any evidence of fusion with the fibula (the tibia and fibula are co-ossified in Basilosaurus) . SAHNI AND MISHRA'S PROTOSIREN AND MOERITHERIID The most distinctive features of GSP-UM 3009 and GSP-UM 3015 are postcranial: a fused sacrum, a pelvis with a deep acetabulum, a massive proximal femur, and a long tibia with a distal articulation for a broad astragalus. Most archaeocetes have similar skulls, and the cranium of archaeocetes is generally not as informative of evolutionary grade as are the vertebral column and appendicular skeleton. Thus, once we referred GSP-UM 3009 and GSP-UM 3015 to the same taxon as the skull GSP-UM 1853, because all come from the upper part of the Domanda Shale and all have similarly shaped vomers, and once we identified the skull as Indocetus ramani, because of its similarity to the type skull (to the extent that these can be compared), it was natural to look for confirming postcranial evidence from the type locality. If Indocetus in Pakistan is distinctive in having a fused sacrum and a pelvis with a deep acetabulum, then Indocetus in India should have the same characteristics. Sahni and Mishra (1975) described five mammalian taxa, three cetaceans, one sirenian, and one proboscidean, from the chocolate-colored Shell (or later Chocolate) Limestone at Harudi in Kutch: (1) the protocetid cetacean Indocetus ramani, based on a partial skull (LUVP 11034, holotype) (2) the protocetid (now remingtonocetid) cetacean Protocetus sloani, based on a mandibular fragment (LUVP 11002, holotype, now placed in Remingtonocetus; Kumar and Sahni, 1986) (3) the protocetid (now remingtonocetid) cetacean Protocetus harudiensis, based on a partial skull with dentaries (LUVP 11037, holotype, now placed in Remingtonocetus; Kumar and Sahni, 1986) (4) the protosirenid sirenian Protosirenfmasi, based on the acetabular region of a pelvis (LUVP 11038), with another bone possibly representing part of the ilium (LUVP 11039) (5) a possible moeritheriid proboscidean, based on a sacrum with two and a one-half vertebral centra fused together (LUVP 11069). As noted above, the partial dentary LUVP 11003 may be from this stratigraphic interval and locality as well. The acetabular region of the Harudi pelvic fragment, LUVP 11038 (see Sahni and Mishra, 1975, fig. 4 and plate VI fig. l), is similar in form, in pattern of breakage, and in size to GSP-UM 3015. Sahni and Mishra (1975, p. 28) gave two measurements of the diameter of the acetabulum: a diameter of 46 mm measured anteroposteriorly, and a diameter of 35 mm measured in some other direction. The average of these two measurements (for a structure that must accommodate a spherical femoral head, being the same diameter in all directions) is 40.5 mm, which compares very closely with the acetabulum diameter of 40 mm in GSP-UM 3015 reported above. Sahni and Mishra stated that the acetabular region of the pelvis is broad, with a width of 52 mm, which compares well with a depth (in life position) of 54 mm in
INDOCETUS RAMANI FROM PAKISTAN
41 1
GSP-UM 3015. After comparison with GSP-UM 3015, there is little doubt that LUVP 10038 is the acetabular region of a pelvis of Indocetus ramani. The bone that suggested sirenian affinity is the isolated element, LUVP 10039, found within a meter of LUVP 10038 and interpreted as a narrow and slender ilium. This does not contact 10039 and it could be a fragment of any of the as yet poorly known long bones of Indocetus. The Harudi sacrum, LUVP 11069 (see Sahni and Mishra, 1975, fig. 5 and plate VI fig. 2), is similar in form, in pattern of breakage, and in size to both GSP-UM 77 and GSP-UM 3009. LUVP 11069 is said to include three fused vertebrae (Sl-S3), which Sahni and Mishra (1975, p. 30) measured as 122 mm in total length (lengths of individual centra are not given). The total length of S1-S3 in GSP-UM 77 and in GSP-UM 3009 is on the order of 168 mm (Table 4). S3 in the Harudi sacrum appears to be much shorter than S1 or S2 in the photograph in Sahni and Mishra's plate VI, fig. 2a, which shows the spacing of sacral foramina, and we suspect that the Harudi sacrum is broken through the middle of S3 (as GSP-UM 77 is broken through the middle of S3). Measurements of the width and height of the centrum of S1 in the Harudi sacrum, measured anteriorly, are slightly smaller than those of known Pakistan specimens, but the height-to-width ratio is similar. Measurements of the height and width of the neural canal are similar. All have similarly developed auricular processes for articulation with the pelvis, similarly developed pleurapophyses, similar-sized sacral foramina, similarly developed neural spines, and similarly developed zygapophyseal pits. Sahni and Mishra (1975, p. 30) wrote in discussing the Harudi sacrum: With the possible exception of Irzdocctus ramani gen. et sp. nov., no other Kutch marine mammal known at present would have possessed such a well-developed sacrum. New specimens from Pakistan show that whales very similar or identical to Indocetus ramani had both (1) a pelvis with a large, deep acetabulum like that from Harudi, and (2) a sacrum composed of multiple fused vertebrae like that from Harudi. Since the type specimen of Indocetus ramani is itself from Harudi, this association is documented there as well. Consequently, there is little doubt that our Pakistan specimens belong to Indocetus ramani, and it seems clear that this genus retained many characteristics of hind limb structure more typical of land mammals, as Sahni and Mishra hinted. DISCUSSION The age, living environment, and functional morphology of Indocetus are all important for interpretation of its evolutionary grade. Age-Many authors have studied the age of lower and middle Khirthar strata in Pakistan, which includes the Domanda Shale yielding Indocetus ramani. Nuttall (1926) regarded this as Lutetian middle Eocene. Biostratigraphic and systematic examination of Eocene mollusks at Rakhi Nala and Zinda Pir (Eames, 1951, 1952a), studied in the context of their distribution elsewhere, convinced Eames (1952b,c) that correlation of the Khirthar Formation or Series of Pakistan with the middle Eocene elsewhere is justified. Since his "lower chocolate clays" or Domanda Shale are in the lower part of the Khirthar Formation or Series, this means that whales from the upper part of the Domanda Shale, like those described here, must be early or middle Lutetian in age. Nagappa (1959) reviewed all of the foraminifera1 evidence and corroborated Eames' placement of the "lower chocolate clays" in the early middle Eocene. Latif (1962) carried out a detailed investigation of pelagic forams at Rakhi Nala. While it is difficult to relate Latif s published stratigraphic section to that of Eames (or to strata in the field), it is clear that Latif placed the lower-to-middle Eocene boundary higher than Eames did (see Latif, 1962, p. 37). The discrepancy may not be meaningful because it concerns a 900 foot thickness of sediments
412
P. D. GINGEFUCH ET AL.
that yielded no planktonic forams (the Domanda Shale?), but the effect, if any, for correlation to the global time scale would be to make the Domanda Shale older than Eames thought. Hemphill and Kidwai (1973) regarded the Domanda Shale as lower middle Eocene. Our best estimate at present is that the Domanda Shale is early Lutetian in age (planktonic foraminiferal zone P10, ca. 48 Ma), an estimate that should become much more precise in the future as this problem is studied using cetacean and foraminiferal biostratigraphy with paleomagnetic reversal stratigraphy in the context of sea level change and sequence stratigraphy. If Indocetus lived 48 million years before present, then it was about 2 m.y. older than Egyptian Protocetus. Depositional environment and taphonomy-Investigations of the Domanda Shale have almost all concentrated on its age. Nuttall (1926, p. 496) interpreted the lower and middle Khirthar series, from the Sulaiman Range in the north, yielding the Indocetus studied here, to Kutch in the south, yielding the Indocetus studied by Sahni and Mishra, to have been deposited in a uniformly "tranquil sea of moderate depth." This is the Sindhu Sea of Sahni and Kumar (1974). The lower part of the Domanda Shale in the Sulaiman Range is predominantly green in color and composed of clay shales. The upper part of the Domanda Shale yielding the Indocetus remains described here includes more-oxidized red and brown clay shales and minor siltstones and sandstones. The unit as a whole has yielded a large marine molluscan fauna (Eames, 1951, 1952a), and remains of other marine organisms including echinoids, small corals, and crustaceans (Gingerich et al., 1979). The only faunal element described from the Domanda Shale and said to be nonmarine is an edentulous maxilla of a "mesonychid" mammal from Safed Tobah south of Rakhi Nala (Pilgrim, 1940). Gingerich et al. (1979) concluded that the maxilla could possibly represent an archaeocete rather than a land mammal, and we are now almost certain of this. There is no evidence that any of the Domanda Shale is nonmarine. The two partial skeletons that form the basis of this report, GSP-UM 3009 and 3015, include bones encrusted on one surface with bivalve spat, which indicates that the skeletons were defleshed and disarticulated, and scattered on the sea bottom for a significant period of time before burial. This may explain why the skeletons are not more complete, and why isolated elements of each were found scattered over an area with a radius of 5-10 meters. Functional morphology-The skull of Indocetus has the characteristic features of an archaeocete, including an elongated rostrum with an anteriorly-protruded palate and dentition; broad flat frontals; large, dense auditory bullae; and a foramen magnum opening more posteriorly than ventrally. The dentition is poorly known, but it is clear from alveoli that the anterior teeth were well spaced in the skull and that the premolars were the largest teeth. Upper molars retained a distinct protocone (Sahni and Mishra, 1975). The postcranial skeleton of Indocetus is still poorly known, but it is clear that Indocetus retained a number of features that are primitive for a cetacean but commonly found in land mammals. The neck is as long as expected for a land mammal like Pachyaena (Fig. 1I ) , and it is not short like that of Egyptian Protocetus (Fig. 14) or later archaeocetes (Kellogg, 1936). The relatively long neck of Indocetus compared to more advanced archaeocetes may represent nothing more than retention of an ancestral condition, but it may also have been advantageous for catching fish or whatever Indocetus fed on. Large hypapophyses on anterior cervical vertebrae of Indocetus indicate the presence of a strong anterior ventral longitudinal ligament in the cervical region, which may have been required to stabilize its long neck while swimming and diving. As noted above, Indocetus has a much longer sacrum than Pachyaena, with longer sacral centra and a minimum of four sacral vertebrae instead of three, and it has a longer proximal caudal region, with longer vertebrae (both absolutely and relatively) than Pachyaena. We interpret these differences as evidence that Indocetus was specialized to some degree as a caudally-driven swimming mammal, which is consistent with recovery from shallow marine shelf paleoenvironments and with moditications of the skull, including enlarged and dense
IWOCETUS RAMANI FROM PAKISTAN
-B
5D
10 q
413
Sacrum
l
c
-
a!
5
Pach yaena ossifraga
" L
c
5
10-
I
0
I
I
10
I
4
20
I
I
30
I
I
40
I
I
50
I
,
60
I
I
70
Vertebral number (position in column)
FIG. 14-Length-of-vertebrae profile of the Egyptian marine middle Eocene archaeocete Protocetus atavus (open circles) compared to length-of-vertebrae profile of the terrestrial North American early Eocene mesonychid Pacllyaena ossifraga (closed-and-open circles; from Zhou et a]., 1992). This shows P. atavus to have been very similar in size to P. osslfraga. Cervical vertebrae of Protocetus are both absolutely and relatively shorter than those of terrestrial Pachyaena, unlike those of Indocetus ramani (Fig. 11). Indocetus had a sacrum with at least four sacral centra and Pachyaena had a sacrum with three sacral centra, while Protocetus had a sacrum with a single sacral centrum. The number of cervical, thoracic, lumbar, and caudal vertebrae in Protocetus is not known with certainty, and no measurable caudals have been found; identifications of vertebrae and measurements plotted here are from Fraas (1904).
auditory bullae. Protocetus has a single sacral centrum, indicating that it had a more flexible lower back and suggesting that much more of the thorax and lumbus, in addition to the tail, was involved in swimming. Finally, Indocetus is distinctive in having large auricular surfaces on the transverse processes of the sacrum, indicating broad articulation of the sacrum with the pelvis. The pelvis has a large and deep acetabulum, the proximal femur is robust, the tibia is long, and there is evidence of a broad tibioastragalar articulation, although the foot of Indocetus has not yet been found. All of these features, taken together, indicate that Indocetus was probably able to support its weight on land, and it was almost certainly amphibious, as early Eocene Pakicetus is interpreted to have been (based on different evidence; Gingerich et al., 1983). We speculate that Indocetus, like Pakicetus, entered the sea to feed on fish, but returned to land to rest and to birth and raise its young. The basicranium of Indocetus, with its large Protocetus-like auditory bullae, shows that Indocetus was more aquatic than Pakicetus, but retention of a sacrum consisting of at least four co-ossified vertebral centra that articulated with a pelvis and long hind limb shows that Indocetus was more terrestrial than Protocetus. Classijication-If there was any doubt before that Indocetus is distinct from Protocetus, this now appears settled: Indocetus is clearly different in retaining a much longer neck, and Indocetus is clearly different in the size and structure of its sacrum and, by inference, its hind limb and tail. These differences have important functional and life history implications, and Indocetus, in addition to being older geologically, appears to have been a more primitive cetacean as well. Indocetus was classified as a protocetid because of cranial resemblances to
Protocetus (Sahni and Mishra, 1975). We do not disagree, but for determining grade of evolution of archaeocetes, skulls are proving to be less informative than the postcranial skeleton. In recognition of its differences, we propose that Indocetus be classified within Protocetidae in a new subfamily Indocetinae. Indocetine protocetids differ from remingtonocetids, dorudontids, and basilosaurids in having Protocetus-like skulls, while retaining long necks, multiple-vertebral sacra, and substantial hind limbs not seen in protocetines. Indocetines were more aquatic than Pakicetus and other pakicetines (Gingerich and Russell, 1990), but at the same time amphibious tail-swimmers less advanced than Protocetus. Protocetines evidently supported very little body weight on their hind limbs, and they were able to use more of their dorsal and ventral musculature to power swimming that involved increased flexion of the lower back and tail. Indocetus represents an important intermediate stage in the early evolution of Cetacea.
ACKNOWLEDGMENTS Our 1977 field party working at Rakhi Nala included Jean-Louis Hartenberger, Mahmoodul Hassan, Kenneth D. Rose, and Donald E. Russell. Specimens collected in 1981 were found by David Bardack, Hassan Shaheed, William J. Ryan, and Neil A. Wells. In 1992 we benefitted greatly from the field experience and stratigraphic expertise of M. Akram Bhatti of the Geological Survey of Pakistan, Lahore. In addition, we thank Ashok Sahni, now of Panjab University, Chandigarh, for access to the type specimen of Indocetus ramani and other LUVP specimens at Lucknow University; Elmar P. J. Heizmann for access to the type specimen of Protocetus atavus and other comparative specimens in the Staatliches Museum fiir Naturkunde, Stuttgart; and John H. Ostrom for the loan of comparative specimens of Pachyaena in the Princeton collection at Yale University, New Haven. We thank Daniel C. Fisher and Mark D. Uhen for help in identifying the vomers, and Gregg F. Gunnel1 and William J. Sanders for reading the text. Bonnie Miljour drew Figure 1 and printed all of the photographs. William J. Sanders prepared the specimens of Indocetus described here. Field research in 1977 and 1981 was funded by the Geological Survey of Pakistan and by the Smithsonian Institution's Foreign Currency Program. Field research in 1992 was funded by the Geological Survey of Pakistan and by the Office of the Vice-President for Research at the University of Michigan.
LITERATURE CITED DEHM, R., and T. OETTINGEN-SPIELBERG. 1958. Paleontologische und geologische Untersuchungen im Tertiar von Pakistan. 2. Die mitteleocanen Saugetiere von Ganda Kas bei Basal in Nordwest-Pakistan. Bayerische Akademie der Wissenschaften, Mathematisch-Naturwissenschaftliche Klasse, Abhandlungen, 91: 1-54. EAMES, F. E. 1951. A contribution to the study of the Eocene in western Pakistan and western India: B. The description of the La~nellibranchiafrom standard sections in the Rakhi Nala and Zinda Pir areas of the western Punjab and in the Kohat District. Philosophical Transactions of the Royal Society, London, B, 235: 31 1-482. 1952a. A contribution to the study of the Eocene in western Pakistan and western India: C. The description of the Scaphopoda and Gastropoda from standard sections int he Rakhi Nala and Zinda Pir areas of the western Punjab and in the Kohat District. Philosophical Transactions of the Royal Society, London, B, 236: 1-168.
.
IhDOCETUS RAMAM FROM PAKISTAN
415
. 1952b. A contribution to the study of the Eocene in western Pakistan and western India: A. The geology of standard sections in the western Punjab and in the Kohat District. Quarterly Journal of the Geological Society, London, 107:159-171. . 1952c. A contribution to the study of the Eocene in western Pakistan and western India: D. Discussion of the faunas of certain standard sections, and their bearing on the classification and correlation of the Eocene in western Pakistan and western India. Quarterly Journal of the Geological Society, London, 107: 173-200. FRAAS, E. 1904. Neue Zeuglodonten aus dem unteren Mitteleocan vom Mokattam bei Cairo. Geologische und Palaeontologische Abhandlungen, Jena, 6: 197-220. GEOLOGICAL SURVEY OF PAKISTAN. 1986. Geo!ogica! map of Dhodii!!. Gen!ngicd Snrvsy ef Pakistan Map Series, 4: 157 (1 sheet). GINGERICH, P. D. 1992. Marine mammals (Cetacea and Sirenia) from the Eocene of Gebel Mokattam and Fayum, Egypt: stratigraphy, age, and paleoenvironments. University of Michigan Papers on Paleontology, 30: 1-84. and D. E. RUSSELL. 1981. Pakicetus inachus, a new archaeocete (Mammalia, Cetacea) from the early-middle Eocene Kuldana Formation of Kohat (Pakistan). Contributions from the Museum of Paleontology, University of Michigan, 25: 235-246. and . 1990. Dentition of early Eocene Pakicetus (Mammalia, Cetacea). Contributions from the Museum of Paleontology, University of Michigan, 28: 1-20. , , D. SIGOGNEAU-RUSSELL, J.-L. HARTENBERGER, S. M. IBRAHIM SHAH, ,M. HASSAN, K. D. ROSE, and R. H. ARDREY. 1979. Reconnaissance survey and vertebrate paleontology of some Paleocene and Eocene formations in Pakistan. Contributions from the Museum of Paleontology, University of Michigan, 25: 105-115. , N. A. WELLS, D. E. RUSSELL, and S. M. I. SHAH. 1983. Origin of whales in epicontinental remnant seas: new evidence from the early Eocene of Pakistan. Science, 220: 403-406. , B. H. SMITH, and E. L. SIMONS. 1990. Hind limbs of Eocene Basilosaurus: evidence of feet in whales. Science, 249: 154-157. HEMPHILL, W. R., and A. H. KIDWAI. 1973. Stratigraphy of the Bannu and Dera Ismail Khan areas, Pakistan. U. S. Geological Survey Professional Paper, 716B: 1-36. KELLOGG, R. 1936. A review of the Archaeoceti. Carnegie Institution of Washington Publications, 482: 1-366. KUMAR, K., and A. SAHNI. 1986. Renzingtonocetus harudiensis, new co~nbination,a middle Eocene archaeocete (Mammalia, Cetacea) from western Kutch, India. Journal of Vertebrate Paleontology, 6: 326-349. LATIF, M. A. 1962. The use of pelagic Foraminifera in the subdivision of the Paleocene-Eocene of the Rakhi Nala, West Pakistan. Geological Bulletin of the Punjab University, Lahore, 2: 31-45. NAGAPPA, Y. 1959. Foraminifera1 biostratigraphy of the Cretaceous-Eocene succession in the IndiaPakistan-Burma region. Micropaleontology, 5: 145-192. NUTTALL, W. L. F. 1926. The zonal distribution of the larger foraminifera of the Eocene of western India. Geological Magazine, London, 63: 495-504. PILGRIM, G. E. 1940. Middle Eocene ~na~nmals from north-west Pakistan. Proceedings of the Zoological Society of London, B, 110: 127-152. QURESHI, M. K. A., S. Q. ABBAS, and M. A. BHATTI. 1987. Stratigraphy of part of Sulaiman Range. Geological Bulletin of the Punjab University, Lahore, 22: 76-91. SAHNI, A., and V. KUMAR. 1974. Palaeogene palaeobiogeography of the Indian subcontinent. Palaeogeography, Palaeoclimatology, Palaeoecology, 15: 209-226. SAHNI, A., and V. P. MISHRA. 1972. A new species of Protocetus (Cetacea) from the middle Eocene of Kutch, western India. Palaeontology, 15: 490-495. and . 1975. Lower Tertiary vertebrates from western India. Monographs of the Palaeontological Society of India, Lucknow, 3: 1-48. SHAH, S. M. I. 1977. Stratigraphy of Pakistan. Memoirs of the Geological Survey of Pakistan, 12: 1-138. THEWISSEN, J. G. M., and S. T. HUSSAIN. 1993. Origin of underwater hearing in whales. Nature, 36 1: 444-445. WELLS, N. A. 1984. Marine and continental sedimentation in the early Cenozoic Kohat Basin and adjacent northwestern Indo-Pakistan. Ph.D. dissertation, University of Michigan, 465 pp.
416
P. D. GINGERICH ET AL.
WEST, R. M. 1980. Middle Eocene large mammal assemblage with Tethyan affinities, Ganda Kas region, Pakistan. Journal of Paleontology, 54: 508-533. ZHOU, X., W. J. SANDERS, and P. D. GINGERICH. 1992. Functional and behavioral implications of vertebral structure in Pachyaena ossifraga (Mammalia, Mesonychia). Contributions from the Museum of Paleontology, University of Michigan, 28: 289-319.
